[hpsdr] Strange AGC Behavior PowerSDR 1.19.3.2, W5WC

[James ( Pete) Gaddie]

I recently installed Power SDR, W5WC Version 1.19.3.2, Date: 12-31-10.
It is used with a 3 board combination, Ozzie, Mercury and Penelope.  The 
computer uses an Intel Core 2 Duo, 2.4GHz, 2GB RAM.

In listening to on the air SSB signals I am noticing some audio 
distortion and non ideal AGC behavior compared to listening to the same 
signal on an analog receiver.  The AGC performance is better than the 
previous version but on my system it still leaves a lot to be desired.

In an effort to see what is happening I ran some test with stepped level 
CW signals injected into Mercury.  The test setup consist of a 
synthesized signal generator with HP 3555B (10dB) and HP 3555C (1dB) 
step attenuators for level stepping. The audio output level of the 
Mercury board was monitored on a Tektronix Oscilloscope.

These test was done on Upper-Sideband with the signal generator offset 
to produce selected tone frequencies of 100Hz to 1 KHz.  The receivers 
AGC-T was set to 90, AGC to Fast, BW 2.4kHz, AGC, AGC Slope 0.0, and 
Hang Threshold at two divisions to the right.  At these settings the AGC 
threshold is at about -100dBm.

The synthesizer was set for an audio tone frequency of 200Hz. If we 
start at -106dBm and step upward in 10dB steps to -16dBm, at levels 
above the AGC threshold the audio output level remains constant as it 
should.  Now we step downward in 10 dB steps from -16dBm.  On the first 
-10dB step the audio level drops by 6dB and hangs at that level for 
ever.  On the next -10 dB step the audio output jumps back up to full 
level where it should be.  We step down two more steps to -56dBm and the 
audio level again drops and hangs 6dB below where it should be.  Down 
one more step, to -66dBm and it jumps back to where it should be.  
Continuing on down to -96dBm the audio level is maintained at 
approximately the correct level.  At higher audio tone frequencies the 
above behavior varies a bit.  As the signal level is stepped the Power 
SDR S-meter always indicates the correct level.

If you set the AGC Hang Threshold to zero (fully left) the audio output 
does not exhibit the above behavior.  At any setting other than zero it 
exhibits this behavior.

Next I did a test of the AGC slope control.  With the Slope Control set 
6 dB, I stepped the signal from below threshold to -16dbm while 
monitoring the audio output level.  Above AGC threshold the audio level 
stayed constant and did not rise by 6dB as it should.  The slope control 
is not working.

I did similar test using PowerSDR-IF Stage Version 1.19.3.5 on a K3 and 
LP-Pan and got similar but not exactly the same results.  Perhaps there 
are some coding mistakes common to both versions of PowerSDR, not just 
the HPSDR version?  It seems unlikely that the above is peculiar only to 
my hardware.

I would be interested if others could run these test to determine if 
they see a similar behavior.

Thanks, 73

Pete W6XX